Head mounted displays (HMDs) enable immersive, virtual reality (VR) and augmented reality (AR), and are revolutionizing human interaction with a variety of applications. In order for the HMDs to correctly render images in accordance with the wearer's current viewing direction, the position and orientation of the head must be tracked in real time with low latency and high accuracy. Traditional motion detection systems, however, require substantial processing of two-dimensional (2D) images on a frame-by-frame basis to find edges and track how the edges are moving. As a result, these systems have very high power and latency limitations.